Large earthquakes not only move the ground, but also adjust the speed of light to Earth’s gravitational field.Now, researchers have Train a computer to recognize these tiny gravitational signalswhich demonstrates how signals can be used to mark the location and magnitude of strong earthquakes almost instantaneously.
Scientists on May 11 nature.
Such a system could help solve a thorny problem in seismology: how to determine the true magnitude of a major earthquake immediately after it hits, says Andrea Licciardi, a geophysicist at the University of the Côte d’Azur in Nice, France. Without this ability, it is difficult to warn of life-saving dangers quickly and effectively.
As a large earthquake breaks, the tremors and tremors send seismic waves through the ground, which appear as large wobbles on seismographs. But current detection methods based on seismic waves are notoriously difficult to distinguish between a magnitude 7.5 and a magnitude 9 earthquake within seconds of such an event.
That’s because the initial magnitude estimate was based on the height of seismic waves called P waves, the first seismic waves to reach a monitoring station. However, for the strongest earthquakes, those initial P-wave amplitudes are the largest, making it difficult to distinguish earthquakes of different sizes.
But seismic waves aren’t the earliest signs of an earthquake. All of this mass that moves in a big earthquake also changes the density of the rock in different locations. These density changes translate into tiny changes in Earth’s gravitational field, producing “elastic gravitational” waves that travel through the ground at the speed of light—even faster than seismic waves.
Those signals were once thought to be too small to be detected, said Martin Vallée, a seismologist at the Paris Institute of Geophysics who was not involved in the new study.Then in 2017, Vallée and his colleagues First person to report seeing these elastic gravity signals in seismic station data. These findings prove, “There’s a window between when an earthquake starts and when you receive it. [seismic] waves,” said Vallée.
But researchers are still thinking about how to translate these elastic gravity signals into an effective early warning system. Because the gravity wobble is so small, it can be difficult to distinguish them from the background noise in the seismic data.When the scientists went back, they found that only six mega-quakes in the past 30 years have produced identifiable elastic gravity signals, including a magnitude 9 Tohokuchong Earthquake In 2011, a devastating tsunami inundated two nuclear power plants in Fukushima, Japan (SN: March 16, 2011). (An initial estimate of the earthquake magnitude based on AP waves is 7.9.)
That’s where computers can come into play, Licciardi said. He and his colleagues created PEGSNet, a machine learning network designed to identify “instant elastic gravitational signals.” The researchers trained the machine using a combination of real earthquake data collected in Japan and 500,000 simulated gravity signals from earthquakes in the same region. Synthetic gravity data is critical for training, Licciardi said, because real data is scarce, and machine learning models need enough input to find patterns in the data.
After the training is complete, the computer will perform a test: tracking the origin and evolution of the 2011 Tohoku earthquake as if it were happening in real time. The results are promising, Licciardi said. The algorithm was able to accurately identify the magnitude and location of an earthquake 5 to 10 seconds earlier than other methods.
Licciardi said the research is proof-of-concept and could potentially serve as the basis for a prototype early warning system. “Right now, it’s tailored to… work in Japan. We want to build something that can work in other regions that are known for strong earthquakes, including Chile and Alaska. Ultimately, the hope is to build something that can work on a global scale. system that works within.
The results suggest that PEGSNet has the potential to be a powerful tool for early earthquake warning, especially when used in conjunction with other earthquake detection tools, Vallée said.
Still, more work needs to be done. On the one hand, the algorithm is trained to find a single point where the earthquake originated, which is a reasonable approximation if you are far away. But up close, the origin of the earthquake is no longer a single point, it’s actually a larger area that has ruptured. Vallée added that if scientists want to accurately estimate where future ruptures will occur, machines need to look for areas, not points.
Further progress is likely to be made in the future as researchers develop more sensitive instruments that can detect disturbances in Earth’s gravitational field caused by smaller earthquakes, while filtering out other sources of background noise that could obscure the signal. Earth is a very noisy environment, from the oceans to the atmosphere, Valle said.
“It’s a bit of the same challenge physicists face when trying to observe gravitational waves,” Vallée said.These Ripples in spacetime, triggered by giant cosmic collisionis a very different type of gravitationally driven wave (SN: 2/11/16). But the gravitational wave signal is also dwarfed by Earth’s noise — in this case, micro-quakes on the ground.